Skip to content
Tutorials Center
My learning

Topics

Lesson 4 of 39 ยท Getting Started

Statistics Basics for SPSS Users: Variables, Levels and Samples

Learn the statistical ideas you need before using SPSS: populations and samples, variable types, nominal, ordinal and scale levels, and how they decide which procedure to choose.

SPSS will compute almost anything you ask for. It will not tell you whether the request makes sense. A tiny amount of statistical vocabulary prevents most beginner mistakes: what a variable is, what level of measurement it has, and how a sample relates to the population you care about. This lesson gives you that foundation, and links it to the Measure setting you will see in Variable View.

Population, sample and statistic

A population is the full group you want to know about: all customers, all patients with a condition, all students in a country. A sample is the part you actually measure. A number computed from a sample is a statistic; the true number for the population is a parameter.

Because samples vary, statistics from two samples differ even if the population is the same. Inferential procedures (t tests, confidence intervals, regression p-values) measure how much a statistic could vary by chance. That is why how the sample was collected matters more than the software. A random sample supports generalizing; a convenience sample of people who happened to answer an online poll often does not.

Python Stand-in: samples vary (seeded simulation)

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Each run of 40 people gives a slightly different mean. The spread of those means is what the standard error in SPSS output describes.

Variables and values

A variable is a characteristic that differs between cases. Its values are the recorded results.

  • Categorical (qualitative): values name groups. Examples: gender, country, yes/no.
  • Numeric (quantitative): values are amounts. Examples: age, income, number of visits.

Numeric variables are discrete when only certain values are possible (a count of 0, 1, 2...) and continuous when any value in a range is possible (height, time).

The four levels of measurement

Stevens' levels describe how much information the numbers carry. SPSS reduces them to three Measure settings.

Level What the numbers mean Example SPSS Measure
Nominal Names only; no order Gender, city, blood type Nominal
Ordinal Order matters; gaps are unknown Satisfaction 1 to 5, education level Ordinal
Interval Equal gaps, no true zero Temperature in Celsius, IQ score Scale
Ratio Equal gaps and a true zero Height, income, age Scale

The level decides which summaries are valid:

Level Sensible center Sensible spread Sensible charts
Nominal Mode None (use frequencies) Bar, pie
Ordinal Median, mode Range, quartiles Bar, boxplot
Scale Mean, median SD, variance, range Histogram, boxplot, scatter

Likert items (strongly disagree to strongly agree) are ordinal in principle. Researchers often treat a sum or mean of many items (a scale) as approximately interval. That is a common and debated practice; for a single item, use medians and nonparametric tests.

Why SPSS asks for the Measure

Several things in SPSS depend on the Measure setting:

  • The Chart Builder offers different chart types depending on measure.
  • Some procedures (Visual Binning, Custom Tables, Chart Builder variable icons) read it.
  • The variable list in dialogs shows an icon: a ruler for scale, a bar chart for ordinal, three circles for nominal. Dialogs such as Crosstabs accept any type, but many (for example Descriptives) only list numeric variables.

A common trap is that SPSS guesses at import. A column of ratings 1 to 5 might be marked Scale because it is numeric. Set it yourself in Variable View under defining variables.

Python Stand-in: which summaries fit this level?

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Independent and dependent variables

Many tests describe a relationship between two kinds of variable:

  • the dependent variable (outcome, response) is what you want to explain;
  • the independent variable (predictor, factor, grouping variable) is what might explain it.

SPSS dialogs use different words for the same idea: Dependent List, Test Variable, Factor, Grouping Variable, Covariate. Learning the pairing is half the battle: for "Does satisfaction differ by gender?", satisfaction is the dependent variable and gender is the independent (grouping) variable. The Role column in Variable View (Input, Target, Both, None, Partition, Split) lets some procedures preselect them, but it is optional for classic tests.

Descriptive and inferential statistics

  • Descriptive statistics summarize the data you have: counts, percentages, means, charts. They make no claim beyond the sample. See frequencies and percentiles.
  • Inferential statistics use the sample to say something about the population: confidence intervals and hypothesis tests. They come in Chapter 5.

A hypothesis test weighs the null hypothesis (no difference, no relationship) against the data. The p-value (SPSS calls it Sig.) is the probability of results at least as extreme as yours if the null hypothesis were true. A small p-value is evidence against it. It is not the probability that the hypothesis is true, and a significant result is not necessarily an important one.

Python Stand-in: a p-value as a simulation

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

This permutation approach is not what SPSS prints for a t test (it uses the t distribution), but the idea of a p-value is the same: how surprising is the result if there were no real difference?

Habits that protect you

  • Decide the research question and the variables before opening SPSS.
  • Check the level of measurement before choosing a statistic.
  • Note the sample size: results from 12 cases are unstable.
  • Remember correlation is not causation; only study design supports causal claims.
  • Report effect sizes and confidence intervals, not only p-values.

Common mistakes

  • Reporting the mean of a nominal code such as region numbers.
  • Treating a significant p-value as proof that an effect is large or true.
  • Comparing groups without checking that the sample sizes and variables make sense.
  • Mixing up dependent and independent variables in a dialog.

Check your understanding

Quick check ยท 5 questions

of 5 answered You scored / 5 ยท perfect!

  1. 1. Which level of measurement fits 'education level: primary, secondary, tertiary'?

    The categories have a natural order but the gaps between them are not known to be equal, so it is ordinal.
  2. 2. Which SPSS Measure setting do interval and ratio variables both receive?

    SPSS merges interval and ratio levels into Scale.
  3. 3. A statistic is computed from:

    A statistic summarizes a sample; a parameter describes the population.
  4. 4. In 'Does average income differ by region?', which is the dependent variable?

    Income is the outcome being explained, region is the grouping (independent) variable.
  5. 5. A p-value of 0.03 means:

    A p-value is a probability about the data under the null hypothesis, not about the hypothesis itself.

Summary

  • A sample is a part of a population; statistics from samples vary by chance.
  • Variables are categorical or numeric; levels of measurement are nominal, ordinal, interval and ratio, with SPSS Measure collapsing the last two into Scale.
  • The level decides which summaries, charts and tests make sense.
  • Identify the dependent and independent variables before opening a dialog.
  • A p-value measures how surprising the data are under the null hypothesis; it does not measure importance.

Next: The Data Editor, Output Viewer and Menus